Optical processor for an artificial neural network

Abstract

An optical processor adapted to emulate an artificial neural network (ANN) having a plurality of interconnected layers, each layer having one or more artificial neurons, the processor having a spatial light modulator (SLM) optically coupled, via an optical mask, to a photodetector array. In one embodiment, the SLM has a plurality of pixels, each pixel being configurable to emulate an output portion of a corresponding artificial neuron in a signal-sending ANN layer. The optical mask has a hologram that encodes the weights corresponding to interlayer connections in the ANN and spatially modulates the light transmitted by the SLM. The photodetectors of the array spatially resolve the interference pattern produced by the spatially modulated light, with each photodetector being configurable to emulate an input portion of a corresponding artificial neuron in a signal-receiving ANN layer.

Description

BACKGROUND OF THE INVENTION[0001]1. Field of the Invention[0002]The present invention relates to artificial neural networks and, more specifically, to an optical processing unit that can be used to emulate such networks.[0003]2. Description of the Related Art[0004]An artificial neural network (ANN) is an information processing paradigm that is inspired by the way biological nerve systems, such as the brain, process information. An information processing system based on this paradigm can be composed of a large number of interconnected processing elements (artificial neurons) working together to solve specific problems. Similar to people, ANNs can learn by example. As such, an ANN can be configured for a specific application, e.g., pattern recognition or data classification, through a learning process, which involves appropriate adjustments of synaptic connections between different neurons and / or decision making in the neuron core.[0005]FIG. 1 schematically shows a simple artificial n...

AI Technical Summary

Benefits of technology

[0011]Problems in the prior art are addressed by various embodiments of an optical processor adapted to emulate an artificial neural network (ANN) having a plurality of interconnected layers, each layer having one or more artificial neurons, the processor having a spatial light modulator (SLM) optically coupled, via an optical mask, to a photodetector array. In one embodiment, the SLM has a plurality of pixels, each pixel being configurable to emulate an output portion of a respective artificial neuron in a signal-sending ANN layer and having an optical window, which is adapted to become either substantially transparent or substantially opaque based on the putative output of the artificial neuron whose output portion is being emulated. In the transparent state, the optical window transmits a laser pulse toward the photodetector array. In the opaque state, the optical window blocks the laser pulse. The optical mask comprises a hologram that encodes the weights corresponding to interlayer connections in the ANN and spatially modulates the light transmitted by the SLM using this hologram. The photodetector array spatially resolves the interference pattern produced by the spatially modulated light, with each photodetector in the array being configurable to emulate an input portion of a corresponding artificial neuron in a signal-receiving ANN layer. Light intensity measured by each photodetector is compared with a respective threshold value representing the firing threshold for the artificial neuron whose input portion is being emulated by the photodetector to determine a putative output for that artificial neuron. Advantageously over neuron-by-neuron processing of computer emulated ANNs, embodiments of an optical processor of the invention can perform layer-by-layer processing of signal propagation in the ANN being emulated, which results in a corresponding increase of the processing speed.

Problems solved by technology

Disadvantageously, this serial processing causes the emulated ANN to have a relatively slow processing speed.

Images